PaLI-3 Vision Language Models: Smaller, Faster, Stronger

This paper presents PaLI-3, a smaller, faster, and stronger vision language model (VLM) that compares favorably to similar models that are 10x larger. As part of arriving at this strong performance, we compare Vision Transformer (ViT) models pretrained using classification objectives to contrastively (SigLIP) pretrained ones. We find that, while slightly underperforming on standard image classification benchmarks, SigLIP-based PaLI shows superior performance across various multimodal benchmarks, especially on localization and visually-situated text understanding. We scale the SigLIP image encoder up to 2 billion parameters, and achieves a new state-of-the-art on multilingual cross-modal retrieval. We hope that PaLI-3, at only 5B parameters, rekindles research on fundamental pieces of complex VLMs, and could fuel a new generation of scaled-up models.

nirogu, avatar

Impressive results! Only wished they had shared some code or any way to replicate the experiments easily

KingsmanVince avatar

indeed it would be great if the authors did so. I personally found some non-official implementations:

KingsmanVince avatar
  • All
  • Subscribed
  • Moderated
  • Favorites
  • machinelearning
  • modclub
  • DreamBathrooms
  • InstantRegret
  • magazineikmin
  • cubers
  • GTA5RPClips
  • thenastyranch
  • Youngstown
  • rosin
  • slotface
  • tacticalgear
  • ethstaker
  • kavyap
  • Durango
  • anitta
  • everett
  • Leos
  • provamag3
  • mdbf
  • ngwrru68w68
  • cisconetworking
  • tester
  • osvaldo12
  • megavids
  • khanakhh
  • normalnudes
  • JUstTest
  • lostlight
  • All magazines